CN109275121A - A kind of Vehicle tracing method based on adaptive extended kalman filtering - Google Patents
A kind of Vehicle tracing method based on adaptive extended kalman filtering Download PDFInfo
- Publication number
- CN109275121A CN109275121A CN201810946889.0A CN201810946889A CN109275121A CN 109275121 A CN109275121 A CN 109275121A CN 201810946889 A CN201810946889 A CN 201810946889A CN 109275121 A CN109275121 A CN 109275121A
- Authority
- CN
- China
- Prior art keywords
- moment
- vehicle
- follows
- roadside unit
- information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/30—Services specially adapted for particular environments, situations or purposes
- H04W4/40—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P]
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/02—Services making use of location information
- H04W4/025—Services making use of location information using location based information parameters
- H04W4/027—Services making use of location information using location based information parameters using movement velocity, acceleration information
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/02—Services making use of location information
- H04W4/029—Location-based management or tracking services
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Traffic Control Systems (AREA)
Abstract
A kind of Vehicle tracing method based on adaptive extended kalman filtering, the following steps are included: first, join in traffic system in intelligent network, the vehicle travelled by DSRC technology automatic identification simultaneously obtains related data, realizes the information exchange of onboard system and roadside unit;Then, for the relevant information of acquisition, quantified first by gun parallax of the quantitative formula to roadside unit and onboard system, acceleration is predicted secondly by autoregressive moving average method, finally carry out trajectory predictions using adaptive extended kalman filtering;Finally, the information handled well is broadcast to roadside unit, in order to next time with the information exchange of onboard system.The Vehicle tracing method based on adaptive extended kalman filtering that the present invention provides a kind of.
Description
Technical field
The invention belongs to field of traffic, join one kind under traffic system more particularly, to intelligent network and are based on adaptive extension
The Vehicle tracing method of Kalman filtering.
Background technique
With the rapid development of our country's economy, the increasingly raising of living standards of the people, it is each that private car initially enters each family
Family facilitates the trip of household well.But vehicle is universal and popular but also urban traffic environment is worsening, out
The traffic behaviors such as unbalanced wagon flow, congested in traffic, tailstock collision, side collision are showed.Along with poor infrastructure and traffic
The quantity of network congestion, road traffic accident increasingly increases, and high traffic accident incidence is sounded the alarm to the whole society, because
This traffic safety is greatly paid close attention to.In recent years, although China carries out road infrastructure and transportation network
Very big improvement, so that traffic accident quantity and the number of casualties are reduced, but total number of accident and incidence are still very high.
Compared with traditional road traffic system, intelligent network connection traffic system is more intended to by " people ", " road ", " vehicle "
And highway communication facility etc. carries out the dynamical system of information exchange.According to being found after a large amount of statistical research in various countries, driver
Fault be to lead to the principal element of traffic accident.Therefore, present road infrastructure cannot perfect situation again
Under, the work that the status information and working process of acquisition other lane vehicles of road are broadcast to current vehicle is very urgent, this
Enable driver preferably to take corresponding remedial measure, reduces driver because of traffic accident caused by making mistakes.
Summary of the invention
In order to overcome, the safety of information transmission mode between existing vehicle is lower, car accident probability of happening is higher
Deficiency, the Vehicle tracing method based on adaptive extended kalman filtering that the present invention provides a kind of.
The technical solution adopted by the present invention to solve the technical problems is:
A kind of Vehicle tracing method based on adaptive extended kalman filtering, the prediction technique include as follows
Step:
1) joining in traffic system in intelligent network, the vehicle travelled by DSRC technology automatic identification simultaneously obtains related data,
Realize the information exchange of onboard system and roadside unit, wherein the step of information exchange is as follows:
Step 1.1: when driving vehicle enters in the range of directional aerial covers, onboard system meeting and roadside unit
Two-way communication is realized by DSRC technology, enables both sides to send the information in itself storage unit simultaneously, wherein vehicle-mounted system
The information that system is sent includes present speed, current location and the timestamp of vehicle, and the information that roadside unit is sent includes other vehicles
Which direction is the predicted position of vehicle on road be located on, a few lanes and acceleration;
Step 1.2: the information of vehicles that roadside unit will acquire is sent to edge Cloud Server and carries out a series of operation
Operation;
2) edge Cloud Server carries out azimuthal angle calculation and does according to the gun parallax between roadside unit and onboard system
Corresponding quantification treatment quantifies vehicle heading for location information, wherein quantizing process are as follows:
Step 2.1: location information is converted into digital information existing for visual angle, wherein the actual bearer angle of vehicle
Relative to roadside unit is defined as:
Here, each parameter definition is as follows:
Azimuth between k moment roadside unit and onboard system;
θk: azimuthal antitrigonometric function is converted by the vehicle location at k moment;
σk: in k moment bearing error noise as caused by signal reflex;
(xk,yk): in the current location of k moment vehicle;
Step 2.2: using crossroad center as coordinate origin, to deflectionQuantification treatment is carried out, determines vehicle institute
Direction, wherein quantitative formula is as follows:
Here, each parameter definition is as follows:
bk: the quantized directions in the moment crossroad k;
Q (): quantization function;
I: the direction signs of crossroad;
Step 2.3: in order to realize that directional information embodies, the lane of the direction being quantified, by practical bearing angleRenamed asIt is recorded as:
Here, each parameter definition is as follows:
Azimuth between k moment roadside unit and onboard system;
θk': the relative position of vehicle and roadside unit is converted azimuthal antitrigonometric function by the k moment;
(xi,yi): the fixation position of roadside unit on the i of direction;
Step 2.4: being directed to above-mentioned quantitative formula, second is carried out to the direction where vehicle and is quantified, determine vehicle place
Lane;
Here, each parameter definition is as follows:
qt: the quantized directions in t moment lane;
ji: j-th of lane on the i of direction;
N: lane sum;
3) assume that p+1 nearest car speed, which is used only, carries out acceleration estimation, p-th of acceleration calculation mode are as follows:
Here, each parameter definition is as follows:
Δ τ: sampling time interval;
Δ v: the difference of the speed of later moment in time and previous moment;
vk-p: in the speed of t-p moment trolley;
τk-p: in the timestamp of t-p moment trolley;
ak-p: p-th of acceleration value;
Hereafter, according to p acceleration value, vehicle acceleration prediction is carried out using autoregressive moving average method, wherein pre-
It is as follows to survey formula:
Here, each parameter definition is as follows:
ak: in the acceleration of k moment trolley;
P: Autoregressive, i.e. acceleration sum;
Q: moving average order, i.e. sliding sum;
β: the undetermined coefficient being not zero;
The undetermined coefficient being not zero;
ξk: in k moment independent error term;
4) assume that driver in the case where not lane change, for the relevant information of acquisition, utilizes adaptive spreading kalman
Filtering algorithm carries out horizontal position prediction to driving vehicle, wherein the calculation formula of horizontal position are as follows:
State equation and observational equation are converted by above-mentioned formula using state-space model, wherein the following institute of equation
Show:
xk=f (xk-1)+Wk-1; (1)
zk=h (xk-1)+Vk-1; (2)
Here, each parameter definition is as follows:
xk: horizontal position value of the vehicle at the k moment;
F (): the real trace function of horizontal position, is nonlinear function;
H (): the observation lopcus function of horizontal position, is nonlinear function;
Wk-1: in the systematic error at k-1 moment, Gaussian distributed N (0, Qk-1), wherein Qk-1=cov (Wk-1), Qk-1
It is in the process-noise variance at k-1 moment;
Vk-1: in the observation error at k-1 moment, Gaussian distributed N (0, Rk-1), wherein Rk-1=cov (Vk-1), Rk-1
It is in the observation noise variance at k-1 moment;
zk: the State Viewpoint measured value of etching system in k;
Hereafter, vehicle horizontal position is carried out using adaptive extended kalman filtering algorithm according to state-space model
Prediction, wherein the step of vehicle location updates is as follows:
Step 4.1: linearization process first being carried out to the nonlinear function in (1) formula and (2) formula, reduces calculation amount and mould
Type complexity, i.e., by nonlinear function f (xk-1) and h (xk-1) in filter forecasting valueFirst order Taylor expansion is done at place, as follows:
Here, each parameter definition is as follows:
Φk|k-1: the systematic state transfer matrix at k-1 moment to k moment, other k here | when k-1 is the k-1 moment to k
It carves;
Hk-1: the observation transfer matrix at k-1 moment;
Δt1: number is higher than primary higher order indefinite small;
Δt2: number is higher than primary higher order indefinite small;
Hereafter, ignore higher order indefinite small, and (3) formula and (5) formula substituted into (1) formula and (2) formula respectively and obtained:
Step 4.2: calculating priori estimates, it is assumed that current state is at the k-1 moment, by the predicted value at k-1 momentAnd real trace function f (xk-1) go to predict the priori estimates x of k moment trolley horizontal positionk|k-1, whereinK-1 | k-1 refers to some value inscribed when calculating k-1 based on the k-1 moment, works as k | k-1 is not
The described meaning is some value inscribed when calculating k based on the k-1 moment when subscript of state-transition matrix;
Step 4.3: prior uncertainty covariance is calculated, by the posteriori error covariance matrix P at k-1 momentk-1|k-1It goes to predict
The prior uncertainty covariance matrix P at current timek|k-1, whereinμk
For adaptive forgetting factor;
Step 4.4: according to current time prior uncertainty covariance matrix Pk|k-1Calculate kalman gain Kk, whereinRk-1For in the observation noise of Gaussian Profile;
Step 4.5: according to the priori estimates x at k momentk|k-1With observation zk, obtain the posterior estimate at k moment
xk|k, wherein xk|k=xk|k-1+Kk(zk-h(xk|k-1));
Step 4.6: updating posteriori error covariance matrix Pk|k, for the prior uncertainty covariance matrix for calculating the k+1 moment
It prepares, wherein Pk|k=(I-KkHk)Pk|k-1, I is unit matrix;
Step 4.7: updating k and be k=k+1 and come back to the calculating that step 4.2 starts a new round;
5) finally, Cloud Server by the information handled well (predetermined speed of vehicle, be located on which direction, a few lanes and
Acceleration) be broadcast to roadside unit, in order to next time with the information exchange of onboard system.
Further, in the step 4.3, the calculation formula of adaptive forgetting factor are as follows:
Gk=d (Mk-Hk-1Qk-1Hk-1-Rk-1);
ek=zk-h(xk|k-1);
Here, each parameter definition is as follows:
Max { }: it is maximized more afterwards;
α: correction factor can force the tracking performance for improving filter;
Gk,ξk: centre derives variable, no actual physical meaning;
ek: new breath, i.e., the difference between k moment true observation and estimation output valve;
U: largest tolerable error;
D: weight factor can be lowered adaptive when new breath is more than largest tolerable error by reducing weight factor
The value of the factor, finally declines error;
Mk: newly cease the covariance matrix inscribed in k.
Further, in the step 1), in intelligent network connection traffic system, roadside unit is installed at the parting of the ways
On traffic lights and side is accompanied with Cloud Server and directional aerial, wherein the launch angle of alignment antenna is 60 °, keeps roadside single
Member preferably can carry out information exchange with the onboard system in vehicle.
Further, in the step 1.2, it is contemplated that the memory capacity of edge Cloud Server is limited, so by server
In data be zeroed out every other week.
Technical concept of the invention are as follows: firstly, passing through DSRC technology automatic identification row in intelligent network connection traffic system
The vehicle sailed simultaneously obtains related data, realizes the information exchange of onboard system and roadside unit.Then, for the correlation of acquisition
Information is quantified using gun parallax of the quantitative formula to roadside unit and onboard system;Utilize autoregressive moving average method pair
Acceleration is predicted;Horizontal position prediction is carried out using adaptive extended kalman filtering.Finally, the information that will be handled well
Be broadcast to roadside unit, in order to next time with the information exchange of onboard system.
Beneficial effects of the present invention are mainly manifested in: 1, by quantization to roadside unit and the gun parallax of onboard system,
Several lanes which direction current vehicle is located at can be apparent from.2, in conjunction with autoregressive moving average method and adaptive expansion
It opens up Kalman filtering method and realizes the prediction of acceleration and position, and send result to driver, so that driver can root
Suitable judgement and behavior are made according to the relevant information and experience of vehicle, effectively reduce traffic accident
Rate.
Detailed description of the invention
Fig. 1 is the schematic diagram of mobile interchange traffic system information exchange.
Specific embodiment
Present invention is further described in detail with reference to the accompanying drawing.
Referring to Fig.1, a kind of Vehicle tracing method based on quantization adaptive extended kalman filtering, base of the present invention
In a kind of Information Interaction Model under DSRC technology communication (as shown in Fig. 1).It is logical first in intelligent network connection traffic system
It crosses quantitative formula to quantify the gun parallax of roadside unit and onboard system, secondly by autoregressive moving average method to adding
Speed is predicted, finally carries out position prediction, the Vehicle tracing method using adaptive extended kalman filtering
The following steps are included:
1) joining in traffic system in intelligent network, the vehicle travelled by DSRC technology automatic identification simultaneously obtains related data,
Realize the information exchange of onboard system and roadside unit, wherein the step of information exchange is as follows:
Step 1.1: when driving vehicle enters in the range of directional aerial covers, onboard system meeting and roadside unit
Two-way communication is realized by DSRC technology, enables both sides to send the information in itself storage unit simultaneously, wherein vehicle-mounted system
The information that system is sent includes present speed, current location and the timestamp of vehicle, and the information that roadside unit is sent includes other vehicles
Which direction is the predicted position of vehicle on road be located on, a few lanes and acceleration;
Step 1.2: the information of vehicles that roadside unit will acquire is sent to edge Cloud Server and carries out a series of operation
Operation, it is contemplated that the memory capacity of edge Cloud Server is limited, so the data in server are zeroed out every other week;
2) edge Cloud Server carries out azimuthal angle calculation and does according to the gun parallax between roadside unit and onboard system
Corresponding quantification treatment quantifies vehicle heading for location information, wherein quantizing process are as follows:
Step 2.1: location information is converted into digital information existing for visual angle, wherein the actual bearer angle of vehicle
Relative to roadside unit is defined as:
Here, each parameter definition is as follows:
Azimuth between k moment roadside unit and onboard system;
θk: azimuthal antitrigonometric function is converted by the vehicle location at k moment;
σk: in k moment bearing error noise as caused by signal reflex;
(xk,yk): in the current location of k moment vehicle;
Step 2.2: using crossroad center as coordinate origin, to deflectionQuantification treatment is carried out, determines vehicle institute
Direction, wherein quantitative formula is as follows:
Here, each parameter definition is as follows:
bk: the quantized directions in the moment crossroad k;
Q (): quantization function;
I: the direction signs of crossroad;
Step 2.3: in order to realize that directional information embodies, the lane of the direction being quantified, by practical bearing angleRenamed asIt is recorded as:
Here, each parameter definition is as follows:
Azimuth between k moment roadside unit and onboard system;
θk': the relative position of vehicle and roadside unit is converted azimuthal antitrigonometric function by the k moment;
(xi,yi): the fixation position of roadside unit on the i of direction;
Step 2.4: being directed to above-mentioned quantitative formula, second is carried out to the direction where vehicle and is quantified, determine vehicle place
Lane;
Here, each parameter definition is as follows:
qt: the quantized directions in t moment lane;
ji: j-th of lane on the i of direction;
N: lane sum;
3) assume that p+1 nearest car speed, which is used only, carries out acceleration estimation, p-th of acceleration calculation mode are as follows:
Here, each parameter definition is as follows:
Δ τ: sampling time interval;
Δ v: the difference of the speed of later moment in time and previous moment;
vk-p: in the speed of t-p moment trolley;
τk-p: in the timestamp of t-p moment trolley;
ak-p: p-th of acceleration value;
Hereafter, according to p acceleration value, vehicle acceleration prediction is carried out using autoregressive moving average method, wherein pre-
It is as follows to survey formula:
Here, each parameter definition is as follows:
ak: in the acceleration of k moment trolley;
P: Autoregressive, i.e. acceleration sum;
Q: moving average order, i.e. sliding sum;
β: the undetermined coefficient being not zero;
The undetermined coefficient being not zero;
ξk: in k moment independent error term;
4) assume that driver in the case where not lane change, for the relevant information of acquisition, utilizes adaptive spreading kalman
Filtering algorithm carries out horizontal position prediction to driving vehicle, wherein the calculation formula of horizontal position are as follows:
State equation and observational equation are converted by above-mentioned formula using state-space model, wherein the following institute of equation
Show:
xk=f (xk-1)+Wk-1; (1)
zk=h (xk-1)+Vk-1; (2)
Here, each parameter definition is as follows:
xk: horizontal position value of the vehicle at the k moment;
F (): the real trace function of horizontal position, is nonlinear function;
H (): the observation lopcus function of horizontal position, is nonlinear function;
Wk-1: in the systematic error at k-1 moment, Gaussian distributed N (0, Qk-1), wherein Qk-1=cov (Wk-1), Qk-1
It is in the process-noise variance at k-1 moment;
Vk-1: in the observation error at k-1 moment, Gaussian distributed N (0, Rk-1), wherein Rk-1=cov (Vk-1), Rk-1
It is in the observation noise variance at k-1 moment;
zk: the State Viewpoint measured value of etching system in k;
Hereafter, vehicle horizontal position is carried out using adaptive extended kalman filtering algorithm according to state-space model
Prediction, wherein the step of vehicle location updates is as follows:
Step 4.1: linearization process first being carried out to the nonlinear function in (1) formula and (2) formula, reduces calculation amount and mould
Type complexity, i.e., by nonlinear function f (xk-1) and h (xk-1) in filter forecasting valueFirst order Taylor expansion is done at place, as follows:
Here, each parameter definition is as follows:
Φk|k-1: the systematic state transfer matrix at k-1 moment to k moment, other k here | when k-1 is the k-1 moment to k
It carves;
Hk-1: the observation transfer matrix at k-1 moment;
Δt1: number is higher than primary higher order indefinite small;
Δt2: number is higher than primary higher order indefinite small;
Hereafter, ignore higher order indefinite small, and (3) formula and (5) formula substituted into (1) formula and (2) formula respectively and obtained:
Step 4.2: calculating priori estimates, it is assumed that current state is at the k-1 moment, by the predicted value at k-1 momentAnd real trace function f (xk-1) go to predict the priori estimates x of k moment trolley horizontal positionk|k-1, whereinK-1 | k-1 refers to some value inscribed when calculating k-1 based on the k-1 moment, works as k | k-1 is not
The described meaning is some value inscribed when calculating k based on the k-1 moment when subscript of state-transition matrix;
Step 4.3: prior uncertainty covariance is calculated, by the posteriori error covariance matrix P at k-1 momentk-1|k-1It goes to predict
The prior uncertainty covariance matrix P at current timek|k-1, whereinμk
For adaptive forgetting factor;
Step 4.4: according to current time prior uncertainty covariance matrix Pk|k-1Calculate kalman gain Kk, whereinRk-1For in the observation noise of Gaussian Profile;
Step 4.5: according to the priori estimates x at k momentk|k-1With observation zk, obtain the posterior estimate at k moment
xk|k, wherein xk|k=xk|k-1+Kk(zk-h(xk|k-1));
Step 4.6: updating posteriori error covariance matrix Pk|k, for the prior uncertainty covariance matrix for calculating the k+1 moment
It prepares, wherein Pk|k=(I-KkHk)Pk|k-1, I is unit matrix;
Step 4.7: updating k and be k=k+1 and come back to the calculating that step 4.2 starts a new round;
5) finally, Cloud Server by the information handled well (predetermined speed of vehicle, be located on which direction, a few lanes and
Acceleration) be broadcast to roadside unit, in order to next time with the information exchange of onboard system.
Further, in the step 4.3, the calculation formula of adaptive forgetting factor are as follows:
Gk=d (Mk-Hk-1Qk-1Hk-1-Rk-1);
ek=zk-h(xk|k-1);
Here, each parameter definition is as follows:
Max { }: it is maximized more afterwards;
α: correction factor can force the tracking performance for improving filter;
Gk,ξk: centre derives variable, no actual physical meaning;
ek: new breath, i.e., the difference between k moment true observation and estimation output valve;
U: largest tolerable error;
D: weight factor can be lowered adaptive when new breath is more than largest tolerable error by reducing weight factor
The value of the factor, finally declines error;
Mk: newly cease the covariance matrix inscribed in k.
Further, in the step 1), in intelligent network connection traffic system, roadside unit is installed at the parting of the ways
On traffic lights and side is accompanied with Cloud Server and directional aerial, wherein the launch angle of alignment antenna is 60 °, keeps roadside single
Member preferably can carry out information exchange with the onboard system in vehicle.
Claims (4)
1. a kind of Vehicle tracing method based on adaptive extended kalman filtering, which is characterized in that the method includes
Following steps:
1) join in traffic system in intelligent network, the vehicle travelled by DSRC technology automatic identification simultaneously obtains related data, realizes
The information exchange of onboard system and roadside unit, wherein the step of information exchange is as follows:
Step 1.1: when driving vehicle enters in the range of directional aerial covers, onboard system can pass through with roadside unit
DSRC technology realizes two-way communication, and both sides is enabled to send the information in itself storage unit simultaneously, wherein onboard system is sent
Information include vehicle present speed, current location and timestamp, roadside unit send information include that other lanes are got on the bus
Predicted position, be located on which direction, a few lanes and acceleration;
Step 1.2: the information of vehicles that roadside unit will acquire is sent to edge Cloud Server and carries out a series of arithmetic operation;
2) edge Cloud Server carries out azimuthal angle calculation and does corresponding according to the gun parallax between roadside unit and onboard system
Quantification treatment quantifies vehicle heading for location information, wherein quantizing process are as follows:
Step 2.1: location information is converted into digital information existing for visual angle, wherein the actual bearer angle of vehicleRelatively
In roadside unit is defined as:
Here, each parameter definition is as follows:
Azimuth between k moment roadside unit and onboard system;
θk: azimuthal antitrigonometric function is converted by the vehicle location at k moment;
σk: in moment k bearing error noise as caused by signal reflex;
(xk,yk): in the current location of moment k vehicle;
Step 2.2: using crossroad center as coordinate origin, to deflectionQuantification treatment is carried out, determines the side where vehicle
To, wherein quantitative formula is as follows:
Here, each parameter definition is as follows:
bk: the quantized directions in moment k crossroad;
Q (): quantization function;
I: the direction signs of crossroad;
Step 2.3: in order to realize that directional information embodies, the lane of the direction being quantified, by practical bearing angleWeight
It is named asIt is recorded as:
Here, each parameter definition is as follows:
Azimuth between moment k roadside unit and onboard system;
θk': the relative position of vehicle and roadside unit is converted azimuthal antitrigonometric function by the k moment;
(xi,yi): the fixation position of roadside unit on the i of direction;
Step 2.4: being directed to above-mentioned quantitative formula, second is carried out to the direction where vehicle and is quantified, determines the vehicle where vehicle
Road;
Here, each parameter definition is as follows:
qt: the quantized directions in t moment lane;
ji: j-th of lane on the i of direction;
N: lane sum;
3) assume that p+1 nearest car speed, which is used only, carries out acceleration estimation, p-th of acceleration calculation mode are as follows:
Here, each parameter definition is as follows:
Δ τ: sampling time interval;
Δ v: the difference of the speed of later moment in time and previous moment;
vk-p: in the speed of t-p moment trolley;
τk-p: in the timestamp of t-p moment trolley;
ak-p: p-th of acceleration value;
Hereafter, according to p acceleration value, vehicle acceleration prediction is carried out using autoregressive moving average method, wherein predictor formula
It is as follows:
Here, each parameter definition is as follows:
ak: in the acceleration of k moment trolley;
P: Autoregressive, i.e. acceleration sum;
Q: moving average order, i.e. sliding sum;
β: the undetermined coefficient being not zero;
The undetermined coefficient being not zero;
ξk: in k moment independent error term;
4) assume that driver in the case where not lane change, for the relevant information of acquisition, utilizes adaptive extended kalman filtering
Algorithm carries out horizontal position prediction to driving vehicle, wherein the calculation formula of horizontal position are as follows:
State equation and observational equation are converted by above-mentioned formula using state-space model, wherein equation is as follows:
xk=f (xk-1)+Wk-1; (1)
zk=h (xk-1)+Vk-1; (2)
Here, each parameter definition is as follows:
xk: horizontal position value of the vehicle at the k moment;
F (): the real trace function of horizontal position, is nonlinear function;
H (): the observation lopcus function of horizontal position, is nonlinear function;
Wk-1: in the systematic error at k-1 moment, Gaussian distributed N (0, Qk-1), wherein
Qk-1=cov (Wk-1), Qk-1It is in the process-noise variance at k-1 moment;
Vk-1: in the observation error at k-1 moment, Gaussian distributed N (0, Rk-1), wherein
Rk-1=cov (Vk-1), Rk-1It is in the observation noise variance at k-1 moment;
zk: the State Viewpoint measured value of etching system in k;
Hereafter, vehicle horizontal position is predicted using adaptive extended kalman filtering algorithm according to state-space model,
Wherein, the step of vehicle location updates is as follows:
Step 4.1: linearization process first being carried out to the nonlinear function in (1) formula and (2) formula, calculation amount is reduced and model is complicated
Degree, i.e., by nonlinear function f (xk-1) and h (xk-1) in filter forecasting valueFirst order Taylor expansion is done at place, as follows:
Here, each parameter definition is as follows:
Φk|k-1: the systematic state transfer matrix at k-1 moment to k moment, other k here | k-1 is the k-1 moment to k moment;
Hk-1: the observation transfer matrix at k-1 moment;
Δt1: number is higher than primary higher order indefinite small;
Δt2: number is higher than primary higher order indefinite small;
Hereafter, ignore higher order indefinite small, and (3) formula and (5) formula substituted into (1) formula and (2) formula respectively and obtained:
Step 4.2: calculating priori estimates, it is assumed that current state is at the k-1 moment, by the predicted value at k-1 momentAnd
Real trace function f (xk-1) go to predict the priori estimates x of k moment trolley horizontal positionk|k-1, whereinK-1 | k-1 refers to some value inscribed when calculating k-1 based on the k-1 moment, works as k | k-1 is not
The described meaning is some value inscribed when calculating k based on the k-1 moment when subscript of state-transition matrix;
Step 4.3: prior uncertainty covariance is calculated, by the posteriori error covariance matrix P at k-1 momentk-1|k-1When going to predict current
The prior uncertainty covariance matrix P at quarterk|k-1, whereinμkIt is adaptive
Forgetting factor;
Step 4.4: according to current time prior uncertainty covariance matrix Pk|k-1Calculate kalman gain Kk, whereinRk-1For in the observation noise of Gaussian Profile;
Step 4.5: according to the priori estimates x at k momentk|k-1With observation zk, obtain the posterior estimate x at k momentk|k,
In, xk|k=xk|k-1+Kk(zk-h(xk|k-1));
Step 4.6: updating posteriori error covariance matrix Pk|k, for calculate the k+1 moment prior uncertainty covariance matrix do standard
It is standby, wherein Pk|k=(I-KkHk)Pk|k-1, I is unit matrix;
Step 4.7: updating k and be k=k+1 and come back to the calculating that step 4.2 starts a new round;
5) finally, the information handled well is broadcast to roadside unit by Cloud Server, in order to next time with the information of onboard system
Interaction.
2. a kind of car speed prediction technique based on quantization adaptive Kalman filter as described in claim 1, feature
It is: in the step 4.3, the calculation formula of adaptive forgetting factor are as follows:
Gk=d (Mk-Hk-1Qk-1Hk-1-Rk-1);
ek=zk-h(xk|k-1);
Here, each parameter definition is as follows:
Max { }: it is maximized more afterwards;
α: correction factor can force the tracking performance for improving filter;
Gk,ξk: centre derives variable, no actual physical meaning;
ek: new breath, i.e., the difference between k moment true observation and estimation output valve;
U: largest tolerable error;
D: weight factor can lower adaptive factor by reducing weight factor when new breath is more than largest tolerable error
Value, finally error is declined;
Mk: newly cease the covariance matrix inscribed in k.
3. a kind of car speed prediction technique based on quantization adaptive Kalman filter as claimed in claim 1 or 2, special
Sign is: on the traffic lights of roadside unit installation at the parting of the ways and other in intelligent network connection traffic system in the step 1)
While being accompanied with Cloud Server and directional aerial, wherein the launch angle of alignment antenna be 60 °, enable roadside unit preferably with
Onboard system in vehicle carries out information exchange.
4. a kind of Vehicle tracing method based on adaptive extended kalman filtering as claimed in claim 1 or 2, special
Sign is, in the step 1.2, it is contemplated that the memory capacity of edge Cloud Server is limited, so the data in server are every
It was zeroed out every one week.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810946889.0A CN109275121B (en) | 2018-08-20 | 2018-08-20 | Vehicle trajectory tracking method based on adaptive extended Kalman filtering |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810946889.0A CN109275121B (en) | 2018-08-20 | 2018-08-20 | Vehicle trajectory tracking method based on adaptive extended Kalman filtering |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109275121A true CN109275121A (en) | 2019-01-25 |
CN109275121B CN109275121B (en) | 2021-08-03 |
Family
ID=65153924
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810946889.0A Active CN109275121B (en) | 2018-08-20 | 2018-08-20 | Vehicle trajectory tracking method based on adaptive extended Kalman filtering |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109275121B (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110264721A (en) * | 2019-07-01 | 2019-09-20 | 北京理工大学 | A kind of urban intersection nearby vehicle trajectory predictions method |
CN110675435A (en) * | 2019-10-08 | 2020-01-10 | 嘉兴智驾科技有限公司 | Based on Kalman filtering and chi2Vehicle track tracking method for detecting smoothing processing |
CN111340855A (en) * | 2020-03-06 | 2020-06-26 | 电子科技大学 | Road moving target detection method based on track prediction |
CN111415531A (en) * | 2019-06-17 | 2020-07-14 | 北京航空航天大学 | Expressway confluence area traffic conflict discrimination method based on vehicle running track prediction |
CN113115230A (en) * | 2021-03-26 | 2021-07-13 | 北京工业大学 | Vehicle broadcast communication control method based on information physical system |
CN113203422A (en) * | 2021-04-14 | 2021-08-03 | 武汉理工大学 | Freight car state inertia parameter joint estimation method based on size measurement device |
CN113504558A (en) * | 2021-07-14 | 2021-10-15 | 北京理工大学 | Ground unmanned vehicle positioning method considering road geometric constraint |
CN114822028A (en) * | 2022-04-25 | 2022-07-29 | 北京宏瓴科技发展有限公司 | Method and device for correcting vehicle running track and computer equipment |
CN117455960A (en) * | 2023-12-22 | 2024-01-26 | 中国人民解放军海军工程大学 | Passive positioning filtering algorithm for airborne photoelectric system to ground under time-varying observation noise condition |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104635231A (en) * | 2015-02-12 | 2015-05-20 | 西安电子科技大学 | Locomotive radar speed measuring method based on adaptive Kalman filter |
CN104880189A (en) * | 2015-05-12 | 2015-09-02 | 西安克拉克通信科技有限公司 | Low-cost tracking anti-jamming method of antenna for satellite communication in motion |
US20170299729A1 (en) * | 2016-04-19 | 2017-10-19 | Clarion Co., Ltd. | Position estimation system and estimation method |
CN107909815A (en) * | 2017-12-07 | 2018-04-13 | 浙江工业大学 | A kind of car speed Forecasting Methodology based on adaptive Kalman filter |
-
2018
- 2018-08-20 CN CN201810946889.0A patent/CN109275121B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104635231A (en) * | 2015-02-12 | 2015-05-20 | 西安电子科技大学 | Locomotive radar speed measuring method based on adaptive Kalman filter |
CN104880189A (en) * | 2015-05-12 | 2015-09-02 | 西安克拉克通信科技有限公司 | Low-cost tracking anti-jamming method of antenna for satellite communication in motion |
US20170299729A1 (en) * | 2016-04-19 | 2017-10-19 | Clarion Co., Ltd. | Position estimation system and estimation method |
CN107909815A (en) * | 2017-12-07 | 2018-04-13 | 浙江工业大学 | A kind of car speed Forecasting Methodology based on adaptive Kalman filter |
Non-Patent Citations (2)
Title |
---|
MARCO LIPPI, MATTEO BERTINI, AND PAOLO FRASCONI: "Short-Term Traffic Flow Forecasting An Experimental Comparison of Time-Series Analysis and Supervised Learning", 《IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS》 * |
季露等: "基于模型融合和特征关联的视频目标跟踪算法", 《计算机技术与发展》 * |
Cited By (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111415531B (en) * | 2019-06-17 | 2021-08-24 | 北京航空航天大学 | Expressway confluence area traffic conflict discrimination method based on travel track prediction |
CN111415531A (en) * | 2019-06-17 | 2020-07-14 | 北京航空航天大学 | Expressway confluence area traffic conflict discrimination method based on vehicle running track prediction |
CN110264721A (en) * | 2019-07-01 | 2019-09-20 | 北京理工大学 | A kind of urban intersection nearby vehicle trajectory predictions method |
CN110675435A (en) * | 2019-10-08 | 2020-01-10 | 嘉兴智驾科技有限公司 | Based on Kalman filtering and chi2Vehicle track tracking method for detecting smoothing processing |
CN110675435B (en) * | 2019-10-08 | 2022-08-02 | 上海智驾汽车科技有限公司 | Vehicle trajectory tracking method based on Kalman filtering and chi 2 detection smoothing processing |
CN111340855A (en) * | 2020-03-06 | 2020-06-26 | 电子科技大学 | Road moving target detection method based on track prediction |
CN113115230B (en) * | 2021-03-26 | 2023-08-18 | 北京工业大学 | Vehicle broadcast communication control method based on information physical system |
CN113115230A (en) * | 2021-03-26 | 2021-07-13 | 北京工业大学 | Vehicle broadcast communication control method based on information physical system |
CN113203422A (en) * | 2021-04-14 | 2021-08-03 | 武汉理工大学 | Freight car state inertia parameter joint estimation method based on size measurement device |
CN113504558A (en) * | 2021-07-14 | 2021-10-15 | 北京理工大学 | Ground unmanned vehicle positioning method considering road geometric constraint |
CN113504558B (en) * | 2021-07-14 | 2024-02-27 | 北京理工大学 | Ground unmanned vehicle positioning method considering road geometric constraint |
CN114822028A (en) * | 2022-04-25 | 2022-07-29 | 北京宏瓴科技发展有限公司 | Method and device for correcting vehicle running track and computer equipment |
CN114822028B (en) * | 2022-04-25 | 2023-10-10 | 北京宏瓴科技发展有限公司 | Correction method and device for vehicle running track and computer equipment |
CN117455960A (en) * | 2023-12-22 | 2024-01-26 | 中国人民解放军海军工程大学 | Passive positioning filtering algorithm for airborne photoelectric system to ground under time-varying observation noise condition |
CN117455960B (en) * | 2023-12-22 | 2024-03-08 | 中国人民解放军海军工程大学 | Passive positioning filtering method for airborne photoelectric system to ground under time-varying observation noise condition |
Also Published As
Publication number | Publication date |
---|---|
CN109275121B (en) | 2021-08-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109275121A (en) | A kind of Vehicle tracing method based on adaptive extended kalman filtering | |
CN109190811A (en) | A kind of car speed tracking based on adaptive extended kalman filtering | |
US10654476B2 (en) | Autonomous vehicle operational management control | |
CN109147390B (en) | Vehicle trajectory tracking method based on quantization adaptive Kalman filtering | |
BR112019016268A2 (en) | operational management of autonomous vehicle which includes operating a partially observable markov decision-making model instance | |
CN109285376A (en) | A kind of bus passenger flow statistical analysis system based on deep learning | |
CN108053645A (en) | A kind of signalized intersections cycle flow estimation method based on track data | |
CN109272745B (en) | Vehicle track prediction method based on deep neural network | |
CN104634352A (en) | Road matching method based on fusion of probe vehicle movement track and electronic map | |
CN103310651A (en) | Bus arrival prediction method based on real-time traffic status information | |
CN109118787B (en) | Vehicle speed prediction method based on deep neural network | |
CN110400461B (en) | Road network change detection method | |
CN103927872A (en) | Method for predicting multi-period travel time distribution based on floating vehicle data | |
CN103258427A (en) | Urban expressway traffic real-time monitoring system and method based on information physical network | |
CN110446160B (en) | Deep learning method for vehicle position estimation based on multipath channel state information | |
CN102087788A (en) | Method for estimating traffic state parameter based on confidence of speed of float car | |
CN109118786A (en) | A kind of car speed prediction technique based on quantization adaptive Kalman filter | |
CN105096642B (en) | Consider the real-time public transport arrival time Forecasting Methodology of gps data delayed impact | |
CN103440422A (en) | Bus behind-schedule recovering method based on arrival time predication with time window | |
CN102708698A (en) | Vehicle optimal-path navigation method based on vehicle internet | |
CN100466010C (en) | Different species traffic information real time integrating method | |
CN106971546B (en) | Road section bus permeability estimation method based on bus GPS data | |
CN105096590B (en) | Traffic information creating method and traffic information generating device | |
CN108039046A (en) | A kind of city intersection pedestrian detection identifying system based on C-V2X | |
CN109308806A (en) | A kind of the traveling detection method and server of the vehicles |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |